9fans - fans of the OS Plan 9 from Bell Labs
 help / color / mirror / Atom feed
* [9fans] c compiler size bug fix
@ 2009-09-15 22:14 erik quanstrom
  0 siblings, 0 replies; only message in thread
From: erik quanstrom @ 2009-09-15 22:14 UTC (permalink / raw)
  To: 9fans

unfortunately i let this one fester until it cost
us some time.  this problem has cropped up
again http://9fans.net/archive/2009/07/305
(it appears that my worry about expandable arrays
was unfounded.)

the problem is that the size of an undefined
struct as the last element of a struct is not
counted because it is not known at the time.
rather than diagnosing an incomplete reference,
the program compiles and prints an unespected
size, "sizeof A = 4 4".  i don't believe it should
compile at all.

the fix is simply this:

/n/dump/2009/0915/sys/src/cmd/cc/dcl.c:541,547 - dcl.c:541,547
  				l->offset = o;
  			} else {
  				if(l->width <= 0)
- 				if(l->down != T)
+ 				if(l->down != T || l->width < 0)
  					if(l->sym)
  						diag(Z, "incomplete structure element: %s",
  							l->sym->name);

thanks to bwc for identifying the problem (again).
patch submitted: /n/sources/patch/incmpltdiag

- erik

---

#include <u.h>
#include <libc.h>

typedef struct A A;
typedef struct B B;

struct A {
	int	a;
	B;
};

struct B {
	uchar	buf[512];
};

void
main(void)
{
	A a;
	B b;

	print("sizeof A = %d %d\n", sizeof(A), sizeof a);
	print("sizeof B = %d %d\n", sizeof(B), sizeof b);
}



^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2009-09-15 22:14 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-09-15 22:14 [9fans] c compiler size bug fix erik quanstrom

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).